SOM-Based Sample Learning Algorithm for Relevance Feedback in CBIR

摘要

Relevance feedback has been shown to be a very effective tool for enhancing retrieval results in text retrieval. In recent years, the relevance feedback scheme has been applied to Content-Based Image Retrieval (CBIR) and effective results have been obtained. However, most of the conventional feedback process has the problem that updating of metric space is hard to understand visually. In this paper, we propose a CBIR algorithm using Self-Organizing Map (SOM) with visual relevance feedback scheme. Then a pre-learning algorithm in the visual relevance feedback is proposed for constructing user-dependent metric space. We show the effectiveness of the proposed technique by subjective evaluation experiments.